A Mixture IRT Analysis of Risky Youth Behavior

Abstract

The study reported in this manuscript used a mixture item response model with data from the Youth Risk Behavior Survey 2009 (N = 16,410) to identify subtypes of adolescents at-risk for engaging in unhealthy behaviors, and to find individual survey items that were most effective at identifying such students within each subtype. The goal of the manuscript is twofold: (1) To demonstrate the utility of the mixture item response theory model for identifying subgroups in the population and for highlighting the use of group specific item response parameters and (2) To identify typologies of adolescents based on their propensity for engaging in sexually and substance use risky behaviors. Results indicate that four classes of youth exist in the population, with differences in risky sexual behaviors and substance use. The first group had a greater propensity to engage in risky sexual behavior, while group 2 was more likely to smoke tobacco and drink alcohol. Group 3 was the most likely to use other substances, such as marijuana, methamphetamine, and other mind altering drugs, and group 4 had the lowest propensity for engaging in any of the sexual or substance use behaviors included in the survey. Finally, individual items were identified for each group that can be most effective at identifying individuals at greatest risk. Further proposed directions of research and the contribution of this analysis to the existing literature are discussed.

A Constrained Confirmatory Mixture IRT Model: Extensions and Estimation of the Saltus model using Mplus

“…Various mixture IRT models have been utilized to examine subjects' heterogeneity in depression/anxiety (Wanders et al, 2016), extreme response styles (H.-Y. Huang, 2016), at-risk unhealthy behaviors (Finch & Pierson, 2011), and item solving strategies (Mislevy & Verhelst, 1990). Other applications of mixture IRT models also exist, e. g., for studying test speededness (Bolt, Cohen, & Wollack, 2002), models' goodness of fit (Rost, 1990), and differential item functioning (Cho & Cohen, 2010;Cohen & Bolt, 2005).…”

Different Approaches to Covariate Inclusion in the Mixture Rasch Model

“…Further, this approach has also been applied to many practical situations. For instance, in psychopathological testing (e.g., Finch & Pierson, 2011;Maij-de Meij, Kelderman, & van der Flier, 2008), it is of great concern for researchers and clinicians to assign subjects to their most likely type of behavior disorders. In such scenarios, mixture IRT models can be used for diagnostic purposes from which an intervention program may be implemented.…”

Item Response Modeling of Multivariate Count Data With Zero Inflation, Maximum Inflation, and Heaping

“…A review of the literature suggests three methodological approaches for solving three distinct problems in measurement: zero-inflated Poisson psychometric models for the analysis of multivariate zero-inflated count data (L. Wang, 2010), latent variable models to account for heaping in univariate count outcomes (H. Wang & Heitjan, 2008) and extreme responding in multivariate Likert-type items (Böckenholt, 2012;Bolt & Johnson, 2009;Bolt & Newton, 2011;De Boeck & Partchev, 2012;ThissenRoe & Thissen, 2013), and mixture IRT models for clinical assessment in heterogeneous populations (Finch & Pierson, 2011;Finkelman et al, 2011;Sawatzky et al, 2012;Wall et al, 2015). All three methods have utility in particular scenarios; however, we are unaware of any existing unifying framework for the analysis of multivariate count outcomes that also accounts for zero inflation, maximum inflation (K inflation), and heaping.…”

“…Mixture IRT has been applied to the analysis of substance use or risky behavior scales, where many of the items may not be applicable to substantial proportions of the population (Finch & Pierson, 2011;Finkelman et al, 2011;Muthen & Asparouhov, 2006;Sawatzky et al, 2012;Wall et al, 2015). Respondents belonging to one of the latent classes-for example, a subgroup of people who abstain from drinking but are nonetheless asked a series of questions relating to symptoms of alcohol dependence-may not engage with the items the same way as other subgroups in the population, and the mixture IRT modeling can help to account for both individual-and group-level differences.…”

“…Simple modifications to a conventional IRT model are not likely to account for the potential subpopulations and individual differences that manifest in Figure 1. This research attempts to address the challenges of modeling multivariate count data with inflation and heaping by combining methodological approaches from three distinct but related literatures: IRT models for multivariate count data (L. Wang, 2010), latent variable models for heaping (H. Wang & Heitjan, 2008) and extreme responding (Böckenholt, 2012;Bolt & Johnson, 2009;De Boeck & Partchev, 2012;Thissen-Roe & Thissen, 2013), and mixture IRT models (Finch & Pierson, 2011;Finkelman, Green, Gruber, & Zaslavsky, 2011;Sawatzky, Ratner, Kopec, & Zumbo, 2012;Wall, …”